DE102004014538A1 - Method for controlling a BiLevel device and BiLevel device - Google Patents

Abstract

This invention relates to a method of controlling a BiLevel device. The method includes repeatedly measuring an airflow, selecting an expiratory threshold, and selecting an inspiratory threshold. It switches to an inspiratory mode if the airflow is greater than the expiratory threshold when compared to the expiratory threshold. It returns to an expiratory mode if the airflow is less than the inspiratory threshold when compared to the inspiratory threshold. Furthermore, the invention relates to a BiLevel device for carrying out the method.

Description

The The invention relates to a method in the preamble of Claim 1 type and a bi-level device, such a method performs. Specifically, this invention relates to how, despite the BiLevel device generated Pressure fluctuations inspiration and expiration can be reliably detected.

BiLevel devices as well the somewhat simpler CPAP devices serve the pneumatic splinting of the respiratory tract to obstructive breathing disorders during the To prevent sleeping.

For the treatment of apneas, continuous positive airway pressure (CPAP) therapy was developed in Chest. Volume No. 110, pages 1077-1088, October 1996 and Sleep, Volume No. 19, pages 184-188. A CPAP device generates by means of a compressor or a turbine a positive overpressure up to about 30 mbar and applies it preferably via a humidifier, a hose and a nasal mask in the respiratory tract of the patient. This overpressure is intended to ensure that the upper respiratory tract remains fully open throughout the night and thus no apnea occurs ( DE 198 49 571 A1 ), which disturb the sleep of the patient. The required overpressure depends, among other things, on the sleep stage and the body position of the sleeping person.

The pressure is often perceived by patients as disturbing. In order to set the overpressure as low as possible but as high as necessary, so-called Auto CPAP devices (cf. 1 ) developed. Algorithms for setting the optimum overpressure are known, for example, from WO 00/24446 A1, WO 02/00283 A1 and WO 02/083221 A1. In addition to air pressure, as with simpler CPAP devices, Auto CPAP devices also measure airflow to the patient. In the processing of the air flow, maxima and minima in the time derivative of the air flow are searched for and determined on the basis of these extremes of inspiration and expiration phases.

One another approach to the pneumatic splint for the patient as possible To make little unpleasant, are so-called bi-level devices. BiLevel devices support the Respiration of the patient by the patient during the Inspiration a little higher Pressure as while the expiration is applied.

Due to the different pressures during inspiration and expiration, it is necessary for BiLevel devices to determine inspiratory and expiratory phases. In the from WO 98/35715 A1 and EP 0 656 216 A2 known BiLevel devices, the time derivative of the air flow with thresholds is compared to distinguish between inspiration and expiratory phases. According to WO 98/35715 A1, this method was common in the prior art (comprehensive paragraph of pages 1 and 2).

It is on BiLevel devices technically difficult the transitions between Capture inspiration and expiration precisely due to air flow because exactly at these transitions the Changed print shall be. By the pressure change The air in the breathing tube and the lungs of the patient slightly compressed or expanded, which causes breathing generated air flow superimposed on a generated by the pressure change air flow becomes. The one by the pressure change generated air flow is now just at the times particularly large, the exact should be recorded.

Problematic is also that a derivative acts like a high-pass filtering and thus leads to a roughening of the signal. This causes noise stronger out. Due to the roughness of the signal, the simple comparison with thresholds lead to incorrect results. That is why in the WO 02/083221 A2 combines the derivative with low-pass filtering and this as "treasuring the Derivative ". On the other hand, low-pass filtering has the disadvantage that it causes the Rise or fall of a signal delayed.

It Object of the invention an improved method and an improved BiLevel device indicate the dates of the transitions between inspiration and Detecting expiration more accurately and thus the inspiratory and expiratory phases determine more precisely.

These The object is achieved by the teaching of the independent claims.

preferred embodiments The invention are the subject of the dependent claims.

Advantageous Comparing the airflow with thresholds is that here a roughening of the air flow through a time derivative as well a delay is avoided by low-pass filtering.

The Choose the minimum air flow during the previous expiratory phase as a threshold immediately after switching to inspiratory mode and selecting the maximum air flow during the previous inspiratory phase as a threshold immediately after switching to an expiratory mode prevents due to the pressure change Fluctuations in the flow of air too quickly become another Switching to the other respiratory mode is done.

Also the extra Compare the derivative of airflow with different ones Thresholds during The inspiratory and expiratory phases prevent an undesirable Switch to the other respiratory mode.

to noise reduction the measured air flow is first subjected to median filtering and then averaged before the derivation the air flow is calculated. The median filtering suppresses in an advantageous manner Way outliers. The combination of median filtering over a few measured values and one subsequent Averaging over the double number of readings provides an optimal compromise between Calculation cost, filter life and required averaging.

The additional consideration the actual pressure increased the reliability switching between inspiratory and expiratory modes. One Aspect here is that when coughing or sneezing is not in the inspiratory mode should be switched. Here, the actual pressure rises above the Target pressure because the pressure control loop is too sluggish to such fast Compensate for pressure fluctuations. The other aspect is that in the Inspiration mode should be switched when the actual pressure is below the target pressure is and the air flow in a third predetermined Time span has risen sharply.

The Lowering the threshold for the air flow during Exhaling shortly after switching to expiratory mode makes the process increasingly sensitive to switching back to inspiratory mode. This corresponds advantageously to the average duration an expiratory phase.

In Accordingly advantageously, the threshold value for the air flow while of inspiration approximately raised in proportion to the current air flow until the current Air flow reaches a maximum. After reaching the maximum the threshold is kept approximately constant.

in the Below is a preferred embodiment of the invention below With reference to the accompanying drawings explained in more detail. Showing:

1 a diagram with the time course of the air flow, the derivation of the air flow and the threshold value for the air flow;

2 the hardware of an AutoCPAP or BiLevel device;

3 a flow chart for the inventive control method for a BiLevel device;

4 a more detailed flowchart for preprocessing;

5 to 7 a more detailed flowchart for expiration processing; and

8th a more detailed flow chart for inspiratory processing.

1 shows a diagram with the time course of the air flow 1 , the derivative of the air flow 2 and the expiratory threshold 3 and the inspiratory threshold 4 , In addition, different time ranges were marked on the below in connection with the flowchart in the 5 to 8th will be discussed in more detail. The expiratory phase was intentionally prolonged by the subject to represent all time ranges provided in the method of the invention.

2 schematically shows the hardware for an AutoCPAP or a BiLevel device 31 , A fan 40 delivers air and provides it under an overpressure of up to 30 mbar. The air is delivered via a breathing tube 32 and a mask 33 a patient 34 applied. Through the opening 35 Air constantly escapes into the environment, allowing exhaled air with a high CO ₂ content through the opening 35 is rinsed out. In device 31 is a pressure sensor 36 for measuring the fan 40 built over pressure built-in. This overpressure is referred to below as the actual pressure. In addition, contains device 31 a flow sensor 37 for detecting the air flow. An example is a heating wire 38 of the flow sensor 37 shown. The from the flow sensor 37 and pressure sensor 36 supplied signals are a microprocessor 39 fed. The microprocessor in turn controls the speed of the fan 40 ,

The construction of device 31 follows the trend in electronics to digitize sensor signals as quickly as possible and then digitally process the signal processing. That of microprocessor 39 finished program contains an inner control loop according to which the speed of the fan is controlled so that the pressure sensor 36 measured actual pressure matches as well as possible with a target pressure. The target pressure is specified by other program parts. If the target pressure is set independently of inspiration and expiration, then it is device 31 to a CPAP device. In addition, if the airflow is evaluated to optimize the target pressure, it is an AutoCPAP device. If different target pressures are given for inspiration and expiration, then it is device 31 to a bilevel device. So it depends only on the other program parts, whether device 31 works as AutoCPAP or as BiLevel device.

3 shows a flowchart of the method according to the invention. According to one embodiment, every 10 ms in step 53 the air flow F _i and in step 54 the actual pressure actual pressure is measured. The index i denotes the current air flow. Since the target pressure is evaluated less intensively than the air flow, an index appeared dispensable at the target pressure. The current time is called t. tn denotes the time at which the next measurement is to take place. In step 51 is waited until the next measurement time tn is reached. In step 52 tn is increased by 10 ms and i by 1.

In step 55 preprocessing takes place more precisely on the basis of 4 is explained. In step 56 a check is made as to whether the method is in an inspiratory or expiratory mode. Then the boolean variable blnspiration is true or false. In expiration mode, the expiratory processing is performed 57 and then in step 51 waited for the next measurement time tn.

In inspiration mode, the inspiration processing becomes 58 and then in step 51 maintained.

The preprocessing 55 is more accurate in 4 shown. First, in step 62 a noise estimate performed. The result is stored in variable NF (for "noise flow"). In a simple case, noise estimation can be done during device development. In this case, a constant value is stored in the variable NF. In a more elaborate embodiment, the standard deviation of the air flow measurements F _i from a certain time interval can be stored in the variable NF. The calculation of the standard deviation may be repeated from time to time or made sliding for each i. Advantageously, a time interval is selected in which the derivative of the air flow is as close as possible to zero. This is the case shortly after the device is turned on or in the middle of an inspiratory or expiratory phase.

in ähnlicher Weise wird ein weiterer (Ansprüche: dritter) gleitender Mittelwert Ave1_i aus den 100 Luftflussmesswerten der letzten Sekunde gemäß Formel (2) in Schritt 64 berechnet:

In step 63 a first moving average Ave5 ₁ (Ave = average) over the last 500 air flow measurements F ₁ according to formula (1) is calculated. This corresponds to an average of the air flow over five seconds.

similarly, another (claims: third) moving average Ave1 _i becomes out of the 100 last second air flow readings according to formula (2) in step 64 calculated:

In step 65 Median filtering is performed over the last ten air flow readings according to formula (3). In a median filtering, the mean value or the arithmetic mean of the two middle values is returned or further processed. Median filtering is more complex than calculating the mean value. However, the result is practically unaffected by outliers, whereas outliers are included in the mean value calculation. F med, i = Median (f I-9 , ..., F i ) (3)

Subsequently, a moving average F _{med, i} over 20 median filtered values F _{med, i} in step 66 calculated. The temporal derivative of the airflow SlopeAve _{i is} estimated from the moving average value F _{med, i} . In a simple case, this can be done by calculating SlopeAve _i as the slope of a straight line through two moving average values F _{med, i} and F _{med, ik} , which are separated by k × 10 ms. For example, k can be 20. In another embodiment, a straight line may be adapted to ten consecutive moving averages F _{med, i} according to formula (4), with the least squares minimized. The slope of this line SlopeAve _i is interpreted as estimated airflow derivative. In another embodiment, the absolute errors can also be minimized.

Finally, in steps 67 and 68 Two further moving average values HighAve _i and LowAve _{i are} calculated using in each case 1500 air flow measured values F _i according to formula (5) or (6), which corresponds to a time span of 15 seconds. What is special about these messages is that for HighAve _i only air flow readings F _i during inspiration and for LowAve _i only air flow readings F _i during expiration are considered. If the variable bInspiration is true, the method according to the invention is in the inspiratory mode, so that the air flow measured values F _{i are used} to calculate HighAve _i . Otherwise, the method is in the expiratory mode so that the air flow measurements F _{i are used} to calculate LowAve _i .

Based on 5 to 7 the expiration processing will be in step 57 explained.

In step 72 a so-called offset Off _{i is} calculated for each index i according to formula (7): off i = (HighAve i - Ave i (500)) / 6 (7)

In another embodiment, the offset Off _i according to formula (7 ') in step 72 be calculated. off i = (HighAve i - LowAve i ) / 10 (7 ')

Subsequently, the memory AveHold = Ave1 _i for three seconds in step 78 set if t _ex > 1 s and F _i > Ave5 _i and F _i = Ave1 _i . The latter three conditions are in steps 73 to 75 checked. It follows that also Ave1 _i > Ave5 _i . Here, t _ex designates the time since the last switch to the expiratory mode. The time at which AveHold = Ave1 _{i is} set is stored in memory t _AH0 (AH = AveHold) in step 79 saved. Is one of the steps 73 to 75 checked conditions not met, so in step 76 Check if the three seconds have already expired. If so, memory AveHold = 0 in step 77 set.

In a further embodiment will be added checked before setting memory AveHold, if this is 0. This causes, memory AveHold remains unchanged after setting.

In yet another embodiment, AveHold only becomes active when switching to the inspiratory mode dus reset, so set = 0.

In 6 is explained as the expiration threshold T _Low , the in 1 with the reference number 3 provided is determined. The time since the last switch to the expiratory mode t _ex will be in steps 81 . 83 . 85 . 87 and 89 divided into five areas that in 1 with the reference number 14 . 15 . 16 . 17 respectively. 18 are provided. In the first area 14 when 0 ≤ t _ex <0.25s, the expiratory threshold T _Low according to formula (8) is set equal to the maximum of the air flow during the previous inspiratory phase InMax, which is in 1 with the reference number 8th is provided. This virtually prevents switching to inspiratory mode during the first 0.25s. T low = InMax if 0 ≤ t ex <0.25s (8)

When 0.25s ≦ t _ex <1s, the expiration threshold T becomes _low during the period 15 lowered approximately linearly according to formula (9). Ave5AtSC is equal to Ave5 _i when switching between inspiratory and expiratory modes, here at t _ex = 0, ie at the time of switching from inspiratory mode to expiratory mode. T low = T Low, Initial · (1 - t ex ) + max (Ave5 i + Off i , Ave5AtSC) · t ex (9) when 0.25s ≤ t _ex <1s

Subsequently, the expiration threshold T _Low according to the following formulas 10 to 12 in steps 86 . 88 and 90 for the time periods 16 . 17 and 18 calculated. T low = max (Ave5 i + Off i , Ave5AtSC) if 1s ≤ t ex <2.5s (10) T low = max (Ave5 i + Off i , AveHold + Off i ) if 2.5s ≤ t ex <7s (11) T low = Ave5 i + Off i if 7s ≤ t ex (12)

Next to the expiration threshold. T _Low is determined according to the following C or JavaScript code (see JavaScript The Comprehensive Reference Works, David Flanagan, translator Ralf Kuhnert et al., O'Reilly, Cologne, ISBN 3-930673-56-8), whether that of parts of the software predetermined target pressure SollDruck has already been achieved and the pressure sensor 36 measured actual pressure remains stable at the target pressure. The code line numbers on the right side of the page do not belong to the code, but are for reference only. The following code will be in step 80 go through every 10 ms.

When switching over from the inspiration mode to the expiratory mode, the target pressure SollDruck is lowered stepwise, but the actual pressure IstDruck initially remains unchanged and then gradually drops to the new target pressure. During this phase, the variable PrintReached = 0. After the test in line 1, at which pressure reaches through "!" is inverted, it is checked in this line in line 2 if actual pressure <= set pressure is what is not yet the case. The actual pressure has fallen to or below the setpoint pressure only when the code has been run through later. Now in code line 2 the pressure reaches is increased, ie = 1 is set. At the next code pass, after the check in code line 1, pressure reaches in code line 4 is incremented. Then the inverted variable PrintStable is checked in line 5. Initially, pressure-stable is 0. For this reason, it is then checked in code line 6 whether actual pressure> = set pressure. If the actual pressure Actual pressure overshoots below the setpoint pressure, this condition is not fulfilled until the end of the overshoot. At the next code pass, the condition! PrintStable is no longer fulfilled, so that line 6 is skipped. Pressure Reached and Pressure Stable are further incremented during the further code runs until the condition Actual Pressure> Target Pressure + Lower Pressure Threshold Stable in code line 7 is fulfilled. Lower pressure threshold Stable is a predetermined value in the range of 0.5 to 1 mbar. This can happen either at a second overshoot, where the actual pressure in turn increases lower pressure threshold steady above the target pressure. A second option is coughing or sneezing. The resulting pressure fluctuations are too fast for the pressure control of the BiLevel device, so they are not corrected. The Variable PressureStable will go to step 95 evaluated to prevent switching to inspiratory mode when coughing or sneezing.

• 1) wenn F_i > T_Low && SlopeAve_i > T_SlopeUp && !(!DruckErreicht & &(IstDruck > IstDruck(–9))) oder
• 2) wenn IstDruck<SollDruck-UntereDruckSchwelleUnstabil && F_i > F_i-9 + NF oder
• 3) wenn DruckStabil && IstDruck<SollDruck-UntereDruckSchwelleStabil && F_i > F_i-9 + NF.

One of the following three conditions must be met to switch from exhalation mode to inspiratory mode:

• 1) if F _i > T _Low && SlopeAve _i > T _SlopeUp &&! (! PressureArea reach &&(ActualPrint> ActualPrint (-9))) or
• 2) if actual pressure <target pressure-lower pressure threshold unstable && F _i > F _i-9 + NF or
• 3) if Pressure Stable && Actual Pressure <Target Pressure Lower Pressure Threshold Stable && F _i > F _i-9 + NF.

These conditions are also in the flow chart in 7 shown. The usual case of switching to inspiratory mode is that the conditions are in steps 91 and 92 are satisfied, so that the air flow F _i greater than the expiration threshold T _Low and the derivative SlopeAve _{i are} greater than the threshold T _SlopeUp . Threshold T _SlopeUp can be set on the BiLevel to adjust the sensitivity of switching to inspiratory mode. Threshold T _SlopeUp is for example between 0 and 6 l / s. The in step 94 Verified Condition (! Pressure Reached && (Actual Pressure> Actual Pressure (-9))) prevents switching to inspiratory mode when coughing or sneezing. When coughing, the actual pressure briefly rises above the target pressure. In this case, line number 9 is set to PrintReach = 0. Actual pressure (-9) indicates the actual pressure 90 ms. -9 is an offset relative to the current index. The 90 ms then result from the fact that an actual pressure is measured every 10 ms. The comparison between the current actual pressure and the actual pressure before 90 ms shows whether there is actually a pressure increase.

Conditions 2 and 3 are similarly structured and are in steps 93 and 95 to 97 checked. According to these conditions, the inspiration mode is switched to when the air flow increases with the strong breath of the patient (F _i > F _i -9 + NF) and at the same time the actual pressure falls below the target pressure. Depending on the Variable Pressure Stable variable, the actual pressure drops below the setpoint pressure in steps 96 and 97 required. Lower Pressure Threshold Unstable and Lower Threshold Stable are predetermined constants.

In the 8th illustrated inspiratory processing is less complicated. There are only two time periods for the time t _in since the last switch to inspiratory mode, namely 0 to 0.25 s and later than 0.25 s in step 102 distinguished. Within 0.25 seconds after switching to inspiratory mode, the inspiration threshold T becomes _high , reference number 4 according to formula (13) equal to the minimum of the air flow during the previous expiratory phase ExMin 7 set. This virtually prevents a return to the expiratory mode within the first 0.25 second. T High = ExMin if 0 ≤ t in <0.25s (13)

THigh = InMax – 2NF wenn InMax ≤ Ave5AtSC (15) After the first 0.25 seconds, in step 104 Checks if the previous maximum air flow during the current inspiratory phase InMax is greater than Ave5AtSC. Ave5AtSC is the moving average over 500 air flow readings at the time of switching between inspiratory and expiratory modes, here from exhalation mode to inspiratory mode. If InMax> Ave5AtSC, the inspiratory threshold T _high becomes as per formula 14 in step 106 calculated. Otherwise, the inspiratory threshold T becomes _high according to formula 15 in step 105 calculated.

T High = InMax - 2NF if InMax ≤ Ave5AtSC (15)

The reference numerals 20 . 21 . 22 and 23 in 1 refer to time periods during an inspiratory phase, which are separated by two local and one absolute maximum of the air flow F _i . The inspiratory threshold T _High is pushed upwards as each maximum is exceeded. As a rule, InMax is> Ave5AtSC, so formula (14) is applied.

The inspiration phase left in 1 is less interesting. Here, the inspiration threshold T _high before reaching the maximum 8th of the air flow in the time period 12 always raised and remains after exceeding the maximum 8th in the time period 13 constant.

Then, before the decision in step 107 the following C code is executed. Like the rest of the inspiration processing, the code will also go through every 10ms. The lines of code 11 to 20 correspond to the above lines of code 1 to 10. It should be noted, however, that the target pressure SollDruck is gradually increased during the transition to the inspiration phase, so that approaches the beginning of the inspiration phase, the actual pressure from below the target pressure. As a result, the comparison operators ">" and "<" are just interchanged.

In Code lines 21 to 23 measure how long the actual pressure is above the Target pressure was. The time will increase in the variable PrintUeberZiel counted. The value of PressureOverGate has to be multiplied by 10 ms, in fact to get the time.

• 1) wenn F_i < T_High && SlopeAve_i < T_SlopeDown oder
• 2) wenn F_i < ¾ Ave5AtSC + ¼ InMax oder
• 3) wenn IstDruck > SollDruck + ObereDruckSchwelle oder
• 4) wenn DruckUeberZiel ≥ 25 & & F_i < T_High

der übliche Fall für das Umschalten in den Exspirationsmodus ist, dass die Bedingungen in Schritten 107 und 108 erfüllt sind, dass also der Luftfluss F_i kleiner als der Inspirationsschwellenwert T_High und die Ableitung SlopeAve_i kleiner als der Schwellenwert T_SlopeDown sind. Schwellenwert T_SlopeDown kann am BiLevel-Gerät eingestellt werden, um die Empfindlichkeit des Umschaltens in den Exspirationsmodus anzupassen. Schwellenwert T_SlopeDown liegt beispielsweise zwischen 0 und –18 l/s.One of the following four conditions must be met to switch from inspiratory mode to expiratory mode:

• 1) if F _i <T _High && SlopeAve _i <T _SlopeDown or
• 2) if F _i <¾ Ave5AtSC + ¼ InMax or
• 3) if actual pressure> set pressure + upper pressure threshold or
• 4) if Print Over Target ≥ 25 && F _i <T _High

the usual case for switching to expiratory mode is that the conditions are in steps 107 and 108 are satisfied, that is, that the air flow F _{i is} smaller than the inspiration threshold T _high and the derivative SlopeAve _i less than the threshold T _SlopeDown . Threshold T _SlopeDown can be set on the BiLevel device to adjust the sensitivity of the switch to expiratory mode. Threshold T _SlopeDown is between 0 and -18 l / s, for example.

The second condition in step 109 is checked, only depends on whether the air flow F _i falls below a predetermined threshold.

Further, according to the third condition, it is switched to the exhalation mode when the actual pressure rises by the upper pressure threshold above the setpoint pressure. This condition will be in step 110 checked.

Finally, according to the fourth condition, it is switched to the exhalation mode when the actual pressure is longer than 0.25 s above the target pressure and at the same time the air flow F _i falls below the inspiration threshold T _High , which in step 111 is checked.

When switching to expiration, the boolean variable bInspiration becomes in step 112 set to false.

The The invention was previously based on preferred embodiments explained in more detail. For a specialist However, it is obvious that various modifications and modifications can be made without departing from the spirit of the invention. Therefore, the scope of protection by the following claims and their equivalents established.

1

airflow

2

time Derivation of the

airflow

3

expiration threshold

4

Inspiration threshold

5

Switch of expiratory

in the inspiration mode

6

Switch from the inspiration

in the expiratory mode

7

minimum of the air flow

during the expiration

8th

maximum of the air flow

during the inspiration

11

0 to 0.25s after

Switch in the

inspiration mode

12

0.25s up to the maximum of

airflow

13

maximum of the air flow up

Switch in the

expiration mode

14

0 up to 0.25s after switching

in the expiratory mode

15

0.25 until 1s after switching

in the expiratory mode

16

1 to 2.5s after switching to

the expiration mode

17

2.5 until 7s after switching to

the expiration mode

18

later than 7 s after switching

in the expiratory mode

19

0 up to 0.25s after switching

in the inspiration mode

20

0.25s until the first local

maximum of the air flow

21

between first and second

local Maximum of

airflow

22

between second local and

absolute Maximum of

airflow

23

between Maximum of

airflow and switching to

the expiration mode

31

AutoCPAP- or bilevel device

32

breathing tube

33

mask

34

patient

35

opening

36

pressure sensor

37

flow sensor

38

heating wire

39

microprocessor

40

Fan

51 to 113

 steps

Claims (15)

Method for controlling a BiLevel device with: repeatedly measuring the air flow ( 1 ) wherein air flow measurements are obtained; Selecting an Expiration Threshold ( 3 ); Choosing an inspiratory threshold ( 4 ); characterized by: repeated comparison ( 91 ) of the air flow with the expiratory threshold ( 3 ) during an expiratory mode ( 14 . 15 . 16 . 17 . 18 . 57 ); Switching ( 5 . 98 ) into an inspiratory mode if the air flow ( 1 ) when compared to the expiratory threshold ( 3 ) greater than the expiratory threshold ( 3 ); repeated comparison ( 107 ) of the air flow ( 1 ) with the inspiratory threshold ( 4 ) during inspiration mode ( 11 . 12 . 13 ; 19 . 20 . 21 . 22 . 23 . 58 ); and switching ( 6 . 112 ) in the expiration mode, if the air flow ( 1 ) when compared to the inspiratory threshold ( 4 ) smaller than the inspiratory threshold ( 4 ). Method according to claim 1, characterized by: storing the minimum air flow ( 7 during the expiratory mode; Choose ( 103 ) of the inspiratory threshold equal to the stored minimum airflow ( 7 ) immediately after switching to inspiratory mode; Storing the maximum air flow ( 8th ) during inspiration mode; and dial ( 81 ) of the expiratory threshold equal to the stored maximum air flow ( 8th ) immediately after switching to the exhalation mode. Method according to one of the preceding claims, characterized by: calculating ( 65 . 66 ) an air flow derivative ( 2 ) by estimating the time derivative of the air flow ( 1 ); To compare ( 92 ) of the airflow discharge ( 2 ) with an expiratory derivation threshold during an expiratory mode; Switching ( 5 ) in the inspiratory mode only if in addition the air flow derivation ( 2 ) is greater than the expiratory derivation threshold; To compare ( 108 ) the airflow derivative with an inspiratory derivative threshold during an inspiratory mode; Switching ( 6 ) into the expiratory mode only if, in addition, the airflow derivative is less than the inspiratory derivative threshold. Method according to claim 3, characterized in that estimating the airflow derivative comprises: repeated calculation ( 65 ) of the median from a first predetermined number of consecutive air flow measurements, wherein median values are obtained; Calculating an average air flow ( 66 ) as an arithmetic mean of a second predetermined number of median values; and calculating the time derivative ( 66 ) of the averaged air flow around the air flow derivative ( 2 ) to obtain. Method according to claim 3 or 4, characterized in that the air flow discharge ( 2 ) is averaged over a first predetermined period of time before comparing it to the expiration and inspiratory derivative thresholds ( 66 ) becomes. Method according to one of the above claims, characterized in that not exhaled mode is changed to inspiratory mode when ( 94 ) the actual pressure has not yet settled to the target pressure and the current actual pressure is above the actual pressure, which was measured before a second predetermined period of time before the current time. Method according to one of the preceding claims, characterized in that the mode of exhalation is switched to the inspiratory mode when the actual pressure ( 36 ) is a first predetermined value below the target pressure ( 96 . 97 ) and the current air flow ( 1 ) is a second predetermined value above the air flow ( 93 ), which is before a third predetermined period of time before the current time. Method according to one of the preceding claims, characterized by the steps: calculating ( 63 ) a first moving average over a third predetermined number of air flow measurements; To calculate ( 67 ) of a second moving average over a fourth predetermined number of air flow measurements measured exclusively during an inspiratory mode. Method according to claim 8, characterized by the steps: lowering ( 84 ) of the expiratory threshold during a sixth period ( 16 . 83 ) shortly after switching to the expiratory mode of a maximum air flow ( 8th ) measured during the previous inspiratory phase, to the sum of the first-factor weighted first moving average plus the second-factor-weighted second moving average, the first factor being slightly less than one and the sum of first and second factors is one; and selecting the expiratory threshold ( 86 ) equal to the sum following the sixth period of time. A method according to claim 9, characterized in that during the sixth time period, the expiration threshold is lowered to only one switching average (84) if the switching average is greater than the sum, and the expiratory threshold during a seventh period ( 16 . 17 . 18 ) after the sixth period ( 15 ) is selected equal to the switching average when the switching average is greater than the sum, the switching average being equal to the first moving average at the time of the previous switch to the exhalation mode. Method according to one of claims 8 to 10, characterized by the steps: calculating ( 64 ) a third moving average over a fifth predetermined number of air readings; To save ( 78 ) of the third moving average during the seventh period for an eighth time period ( 76 ), which is shorter than the seventh time period, in a memory when both the air flow and the third moving average are greater than the first moving average (74) and the air flow is equal to the third moving average (75); otherwise save from zero ( 77 ) in the memory; Calculate the difference ( 72 ) from the second moving average minus the first moving average; Multiply ( 72 ) the difference with a third factor to obtain a product; and adding ( 88 ) of the product with the value in memory to obtain the expiration threshold. Method according to claim 10 or claim 11, as far as claim 11 is based on claim 10, characterized in that the inspiration threshold value ( 4 ) during a period of time ( 12 . 13 ; 20 . 21 . 22 . 23 ) is selected to be equal to the fourth-multiplied maximum air flow during the current inspiratory phase plus the fifth-multiplied switch average (106), shortly after switching to the inspiratory mode until the end of the inspiratory mode, the fourth factor being slightly less than one and the sum of the fourth factor plus the fifth factor is equal to one. Method according to one of the preceding claims, characterized in that the inspiration threshold ( 4 ) during a period of time ( 12 . 13 ; 20 . 21 . 22 . 23 ) is selected to be equal to the difference between the maximum of the airflow during the current inspiratory phase minus a third predetermined value shortly after switching to the inspiratory mode until the end of the inspiratory mode ( 105 ). Method according to claims 12 and 13, characterized in that the inspiratory threshold value during the period ( 12 . 13 ; 20 . 21 . 22 . 23 ) according to claim 12, if the maximum of the air flow during the current inspiration phase is greater than the switching mean value ( 104 ) and otherwise the inspiratory threshold during that period ( 12 . 13 ; 20 . 21 . 22 . 23 ) is selected according to claim 13. Ventilator with: a fan ( 40 ); a pressure sensor ( 36 ) for determining the overpressure, under the air from the fan ( 40 ) is delivered; a flow sensor ( 38 ) for measuring the air flow ( 1 ); a processor ( 39 ) connected to the pressure sensor ( 36 ) and the flow sensor ( 38 ) and a pressure signal from the pressure sensor ( 36 ) and an air flow signal from the flow sensor ( 38 ), wherein the processor ( 39 ) stores a sequence of instructions so that the processor ( 39 ) performs in operation a method according to any one of the preceding claims.